Intel has announced the Intel GPU D-Series, the AI GPU formerly code-named Crescent Island.

Built on the Xe3P architecture, it offers 1.3 petaflops of FP8 compute and up to 480 GB of LPDDR5X memory on partner cards, while its 350-watt, air-cooled cards can run models of up to 300 billion parameters on a single card and more than a trillion on an eight-GPU system. Designed for large language models and long-context inference, the GPU is now sampling with customers, though Intel hasn't announced a general availability date.